
Import chats & photos, generate Relationship Memory + Persona digital twin
Ex-skill is an interactive, memory-archiving and conversational digital twin simulation skill on EasyClaw. It allows users to securely import chat logs and photos related to a past partner or significant relationship, structuring them into a persistent Relationship Memory Bank. Based on this memory, the skill can synthesize a conversational "Digital Twin" that mirrors their unique communication style, vocabulary habits, typical emotional expressions, and relationship perspectives.
The skill is built for individuals seeking creative relationship closure, writers developing realistic, dialogue-driven character models, and researchers exploring conversational AI digital twins.
The expected outcome is a highly structured, dual-layer Digital Persona Profile: a secure local memory database detailing shared memories and relationship milestones, paired with an interactive conversational toggle that lets you discuss relationship histories using their cloned communication style.
1. Source data ingestion. Provide public or authorized text-based relationship materials — such as past chat histories, shared emails, or journal notes.
2. Dual-layer persona extraction. The skill parses the ingested data across two distinct layers:
- *Relationship Memory Bank (Facts):* Extracts shared experiences, milestones, key dates, favorite places, and documented relationship arguments.
- *Digital Twin Persona (Style):* Analyzes linguistic patterns — including typical reply lengths, favorite phrases, conversational habits, and emotional expression markers (such as how they expressed anger or affection).
3. Structured database synthesis. The skill organizes the extracted information into a secure, categorized local JSON directory, separating factual memory indices from the style model guidelines.
4. Interactive conversational toggle. It maps the synthesized persona into an interactive chat interface, allowing you to ask questions about past shared experiences or simulate conversational exchanges in their cloned style.
5. Continuous memory appending. You can continuously upload new photo metadata, letters, or chat logs to update specific ex-profiles, ensuring the memory bank remains accurate as you compile materials.
- Dual-layer cloning: Separates relationship facts (Memory Bank) from communication style (Digital Twin).
- Linguistic style analyzer: Replicates specific text patterns, vocabulary habits, and emotional signals.
- Continuous database appending: Update existing profiles by adding new chat logs or photo records.
- Frictionless query interface: Ask questions about past events and get answers rendered in their cloned voice.
- Missing-context prompt: Proactively prompts you for the profile alias or name if queries are ambiguous.
- Local file system storage: Keeps all private relationship data, chat logs, and cloned profiles stored locally.
1. Creating a digital twin for relationship closure
An individual wants to process a past relationship for creative writing or personal closure. They import a text file of selected past chat logs into Ex-skill. It builds a Digital Twin that captures their communication style — allowing the user to have a final, mock conversational exchange to process lingering questions, helping them find emotional closure headlessly.
2. Compiling a relationship memory bank
A user wants to archive their "first love" relationship memories before deleting old chat apps. They provide the chat logs and notes. The skill compiles a clean, structured chronological timeline of their relationship milestones, favorite travel destinations, and key dates, creating a secure, searchable local relationship archive.
3. Developing realistic dialogue for characters in a novel
An author is writing a novel and wants a character's dialog to sound like a realistic, specific archetype (such as a highly analytical, emotionally reserved ex-partner). They feed representative, anonymized text logs into the skill, creating a cloned persona they can interview to generate natural, authentic character dialogue for their drafts.
4. Resolving ambiguous update requests
A user wants to upload more photos to their saved "Ex" profile but types a vague instruction: *"found more photos."* The skill's context logic detects the missing link, checks current profiles, and prompts politely: *"Please specify: which ex profile alias would you like to update?"* preventing data overwrite errors.
5. Simulating emotional behavior patterns for research
A psychology researcher wants to study how different communication styles express conflict. They clone a specific conversational style and ask: *"How did you react when we had our first major argument?"* The cloned skill generates an authentic, style-aligned simulated response based strictly on the logged relationship history.
A user wants to archive their first love's chat logs and create a conversational digital twin for personal closure.
1. They open EasyClaw and activate Ex-skill.
2. They run: *"Create a digital twin based on these chat logs. Let's call her 'First Love'."*
3. The skill parses the texts, extracts the core memory milestones, and models her warm, slightly sarcastic texting style.
4. It outputs the structured Digital Persona Profile and initializes the local database.
5. The user asks: *"What was our favorite restaurant we went to in Seattle?"*
6. The cloned skill scans the memory bank, identifies the event, and replies in her exact voice: *"As if you'd forget! It was that tiny Italian place near Pike Place with the squeaky chairs."*
Digital twin creation and memory extraction cycle completed in under 45 seconds.
Create ex: distill first love's chat history into a skill -> Generate conversational digital twin
Add memory: found more photos -> Missing ex alias, please specify: which ex to update?
Talk about them: what were they like when angry -> Simulate authentic response based on memory bank
Yes. All text parsing, memory extraction, style modeling, and profile writes are executed entirely locally on your machine over secure, encrypted links. Your chat logs, photos, and digital twins are strictly private and never shared.
The style analyzer parses structural chat patterns: average sentence length, punctuation frequency, vocabulary complexity, use of slang or emojis, and typical response delay models, mapping them to a conversational profile.
No. The skill handles the complex text parsing, database structuring, and persona mapping automatically. You simply upload your text logs, and the conversational interface guides you through the setup.
The Memory Bank stores factual events, dates, locations, and milestones extracted from your files. The Digital Twin is the behavioral style model that allows the AI to communicate using their specific voice and vocabulary habits.
The current version of the skill compiles and indexes photo metadata (such as dates, locations, and descriptions) into the chronological Relationship Memory Bank, creating a visually descriptive timeline of your shared experiences.
Yes. Each relationship history is saved under a distinct, isolated profile JSON file in your local workspace. You can switch between profiles by specifying which alias you want to load.
Yes. All profile databases, text logs, and access models are stored locally on your machine. You can completely erase any ex-profile, its JSON data, and associated archives in one click.
If you trigger an action without specifying which profile to update (e.g., typing "add a memory"), the skill's context logic will politely ask you to clarify which ex alias you want to modify, preventing database conflicts.
Yes. You can import chat logs in Chinese, English, German, French, or Spanish, and the skill will parse the language patterns and generate the digital twin in the original language.
Yes. You can ask the digital twin: "write a letter explaining your perspective on our breakup based on our shared history," and it will output a poignant, style-aligned draft reflecting their modeled perspective.
Browse more in General Tools or all skills.
Get EasyClaw, add this skill, and start building AI agent workflows in minutes.
Get EasyClaw Free →